Data are still the same:
## [1] 181 585
First of all, there does not seem to be any reversed item. I am also glad to see that this is a more-or-less homogeneous instrument given the positive (r>0.1) correlations between items that are designed to measure the same concept (symtom severity).
library(psych)
SCL_scores$alpha
## som ocd hos pho dep psy anx par int SUM
## alpha 0.9121688 0.870248 0.7845488 0.8495714 0.9045754 0.7591232 0.865439 0.7997468 0.863819 0.9754223
alpha(scl_items)[1]
## $total
## raw_alpha std.alpha G6(smc) average_r S/N ase mean sd
## 0.9759163 0.9759144 0.9917546 0.3104432 40.51861 0.002505955 1.851709 0.8547099
Cronbach alpha for the total is acceptable (0.97), scale alphas range between 0.76 and 0.91. Items correlate with each other at a level of r=0.31 on average, and the overall mean (1.85) is centered around the mean of response alternatives (0-4).
STAXI is more heterogenous than SCL-90. There are items which correlate negatively with one another, but according to the initial publication, there are no reversely coded items. You can also clearly see the two “squares” in the matrix: this is an indicaton that there are at least two distinct factors among the items (and indeed this is the case, although there are many more factors, not just 2).
library(psych)
STAXI_scores$alpha
## state expr ang_fe ang_ve ang_ph ang_te ang_re ang_expr ang_rep ang_expr_ctrl
## alpha 0.9355913 0.7735135 0.870971 0.9133927 0.8694831 0.9214688 0.755572 0.8308513 0.7496461 0.8519647
## ang_rep_ctrl
## alpha 0.8877338
alpha(staxi_items)[1]
## Some items ( STAX26 STAX28 STAX29 STAX30 STAX32 STAX34 STAX36 STAX38 STAX40 STAX41 STAX42 STAX44 STAX45 STAX46 STAX48 STAX50 STAX52 STAX53 STAX54 STAX56 ) were negatively correlated with the total scale and
## probably should be reversed.
## To do this, run the function again with the 'check.keys=TRUE' option
## $total
## raw_alpha std.alpha G6(smc) average_r S/N ase mean sd
## 0.8795701 0.8850522 0.9626936 0.1190054 7.699602 0.0127537 2.155471 0.3541051
Cronbach alphas are all the the acceptable level (all >= 0.75), but the mean correlation among items is “only” 0.12, because of the many negative correlational coefficients. The scale is by no means suitable as a “total” anger measure, but it wasn’t designed to be used as total either. When we have a closer look at the negative correlations, it becomes evident that “the expression of anger” items are the ones which correlate negatively with “anger repression” items, which makes sense.
DES items are homogeneous as well, which “makes sense”, given that the instrument was developped to measure dissociation (on 3 separate but related dimensions).
No reversed items based on visual inspection.
library(psych)
DES_scores$alpha
## amn abs depers
## alpha 0.8484171 0.8536298 0.7799651
alpha(des_items)[1]
## $total
## raw_alpha std.alpha G6(smc) average_r S/N ase mean sd
## 0.9373249 0.9407478 0.960061 0.3618525 15.87701 0.006534376 17.11208 14.75302
Cronbach alphas are again excellent (>0.78), with the mean correlation around 0.36.